Host custom MCP servers using Databricks apps

2025-08-08

Important

This feature is in Beta.

Host your own custom or third-party MCP servers as Databricks apps. Custom MCP servers are useful if you already have an MCP server you want to deploy or if you want to run a third-party MCP server as a source of tools.

Requirements

An MCP server hosted as a Databricks app must implement an HTTP-compatible transport, such as the streamable HTTP transport.

Host an MCP server as a Databricks app

See the custom MCP server repo for an example of writing your own MCP server and deploying it as a Databricks app.

To host an existing Python MCP server as a Databricks app, follow these steps:

Set up your environment

Use OAuth to authenticate to your workspace. Run the following in a local terminal:
```
databricks auth login --host https://<your-workspace-hostname>
```

Set up the MCP server

Add a requirements.txt to your server's root directory and specify Python dependencies for your server.

Python MCP servers often use uv for package management. If you use uv, add uv and it will handle installing additional dependencies.
Add an app.yaml specifying the CLI command to run your server.

By default, Databricks apps listen on port 8000. If your server listens on a different port, set it using an environment variable override in the app.yaml file.

Example app.yaml:
```
command: [
    'uv',
    'run',
    'your-server-name',
    ..., # optionally include additional parameters here
  ]
```

Upload the MCP server as a Databricks app

Create a Databricks app to host your MCP server:
```
databricks apps create mcp-my-custom-server
```

Upload the source code to Databricks and deploy the app by running the following commands from the directory containing your app.yaml file:

DATABRICKS_USERNAME=$(databricks current-user me | jq -r .userName)
databricks sync . "/Users/$DATABRICKS_USERNAME/mcp-my-custom-server"
databricks apps deploy mcp-my-custom-server --source-code-path "/Workspace/Users/$DATABRICKS_USERNAME/mcp-my-custom-server"

Connect to the custom MCP server

Click the tabs to see how to connect to an MCP server from various environments.

Local environment

Use OAuth to authenticate to your workspace. Run the following in a local terminal:

databricks auth login --host https://<your-workspace-hostname>

Use the authenticated profile from the previous step to access the hosted Databricks app:

from databricks_mcp import DatabricksOAuthClientProvider
from databricks.sdk import WorkspaceClient
from mcp.client.session import ClientSession
from mcp.client.streamable_http import streamablehttp_client

databricks_cli_profile = "DEFAULT"
workspace_client = WorkspaceClient(profile=databricks_cli_profile)

# Replace with your actual custom MCP server URL
mcp_server_url = "https://<workspace-hostname>/serving-endpoints/mcp-my-custom-server/invocations"

async def test_connection_to_server():
   async with streamablehttp_client(
      f"{mcp_server_url}", auth=DatabricksOAuthClientProvider(workspace_client)
   ) as (read_stream, write_stream, _), ClientSession(
      read_stream, write_stream
   ) as session:
      # List available tools
      tools = await session.list_tools()
      print(f"Available tools: {[tool.name for tool in tools.tools]}")

Notebook: service principal

Use a service principal to access the hosted Databricks app in a Databricks notebook.

import json
from databricks.sdk import WorkspaceClient
from databricks_mcp import DatabricksOAuthClientProvider
from databricks.sdk.credentials_provider import ModelServingUserCredentials
from mcp.client.streamable_http import streamablehttp_client as connect
from mcp import ClientSession

async def main():
    app_url = "https://mcp-server.databricksapps.com/mcp"
    client = WorkspaceClient(
      host="https://workspace_url",
      client_id="client_id",
      client_secret="client_secret"
    )

    async with connect(app_url, auth=DatabricksOAuthClientProvider(client)) as (
        read_stream,
        write_stream,
        _,
    ):
        async with ClientSession(read_stream, write_stream) as session:
            await session.initialize()
            tools = await session.list_tools()
            print(f"Available tools: {[tool.name for tool in tools.tools]}")

await main()

Agent code: on-behalf of user

Set up on behalf of user authentication. See Deploy an agent using on-behalf-of-user authentication.

Enable on-behalf of access using ModelServingUserCredentials to access the hosted Databricks app from agent:

import json
from databricks.sdk import WorkspaceClient
from databricks_mcp import DatabricksOAuthClientProvider
from databricks.sdk.credentials_provider import ModelServingUserCredentials
from mcp.client.streamable_http import streamablehttp_client as connect
from mcp import ClientSession

async def main():
    app_url = "https://mcp-server.databricksapps.com/mcp"
    client = WorkspaceClient(credentials_strategy = ModelServingUserCredentials())

    async with connect(app_url, auth=DatabricksOAuthClientProvider(client)) as (
        read_stream,
        write_stream,
        _,
    ):
        async with ClientSession(read_stream, write_stream) as session:
            await session.initialize()
            tools = await session.list_tools()
            print(f"Available tools: {[tool.name for tool in tools.tools]}")

await main()

Log the agent model using apps.apps scope. See On-behalf-of-user authentication.

Agent code: service principal

Enable System auth using service principal to access the hosted Databricks app from agent:

import json
from databricks.sdk import WorkspaceClient
from databricks_mcp import DatabricksOAuthClientProvider
from mcp.client.streamable_http import streamablehttp_client as connect
from mcp import ClientSession

async def main():
    app_url = "https://mcp-server.databricksapps.com/mcp"
    client = WorkspaceClient()

    async with connect(app_url, auth=DatabricksOAuthClientProvider(client)) as (
        read_stream,
        write_stream,
        _,
    ):
        async with ClientSession(read_stream, write_stream) as session:
            await session.initialize()
            tools = await session.list_tools()
            print(f"Available tools: {[tool.name for tool in tools.tools]}")

await main()

Log the agent model using DatabricksApps as a resource. See Specify resources for automatic authentication passthrough (system authentication).

Deploy an agent

The following notebooks show how to author LangGraph and OpenAI agents that call MCP tools.

LangGraph MCP tool-calling agent

Get notebook

OpenAI MCP tool-calling agent